Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Method and system for building image database

A database and image technology, applied in the field of image processing, can solve the problems of small database scale, labor-intensive, slow speed, etc.

Active Publication Date: 2012-12-19
TSINGHUA UNIV
View PDF4 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] In recent years, in order to solve this problem, researchers in the field of computer graphics and multimedia have done research on a large number of Internet image retrieval based on content. For example, a sketch-based Internet image retrieval method and a A cross-type image retrieval method, but none of these methods can retrieve the image content required by the user in real time, so building a large number of image databases with fine contour segmentation annotations for each type of object has become a solution to this problem. However, the existing databases are usually constructed manually, which is slow and labor-intensive, and the resulting database is small in size.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Method and system for building image database
  • Method and system for building image database

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0023] Such as figure 1 Shown, a kind of image database semi-automatic construction method that the present invention proposes is realized by the following steps: A, utilize keyword to download image automatically from Internet; B, carry out object detection to the downloaded image and carry out initial filtering; Carry out fine contour segmentation of the objects to obtain segmentation primitives; D. Filter the segmentation primitives based on contour matching to form an image database; wherein, the segmented foreground objects in each image are called segmentation primitives.

[0024] Preferably, in the step B, when detecting the object conforming to the keyword description, the corresponding foreground object detection or salient area detection algorithm in computer vision is used for detection.

[0025] Preferably, in the step B, the probability of detecting an object conforming to the category keyword needs to be greater than 0.3.

[0026] Preferably, in the step B, the ...

Embodiment 2

[0034] Such as figure 1 As shown, for step A, in order to download high-quality image data from the Internet, first assign a keyword for each type of object. initial correct rate, such as "running man", "jumping dog", "red car"; then put these keywords into Internet image search engines (such as Google, Flickr, Baidu) for keyword search, every A maximum of 3,000 images are kept in a query result; each image has an initial keyword label, and in the next few steps, images that do not contain the object referred to by the keyword in the initial image set will be removed, and the The objects referred to by the keywords are subjected to fine contour segmentation.

[0035] For step B, for each downloaded image, the present invention uses a foreground object detection algorithm in computer vision to detect the general area of ​​the object in the image that matches the keyword description. For example, for images such as "running man", use human body detection algorithm; for "jumpin...

Embodiment 3

[0039] Such as figure 2 As shown, the present invention also provides a processing system for constructing an image database, including: an image download module, an image processing module and a final filter module; the image download module is used to automatically download images from the Internet using keywords; image processing The module is to detect and preliminarily filter the downloaded pictures, and then perform fine contour segmentation to obtain segmentation primitives; the final filtering module performs contour matching-based filtering on the segmentation primitives to finally generate an image database.

[0040] As can be seen from the above embodiments, the present invention divides and filters massive images downloaded from the Internet through image processing and computer vision technology, making it an image database containing object category annotations and fine segmentation annotations. The image database can usually reach With an annotation accuracy ra...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a method and a system for building an image database. The method comprises the following steps of utilizing keywords to automatically download images from the Internet, carrying out object detection on the downloaded images, carrying out preliminary filtering, carrying out fine contour segmentation on detected objects to obtain segmentation pixels, and filtering the obtained segmentation pixels based on contour match to form the image database. The database can generally achieve 80-90 percent of tagging accuracy, and a user can quickly carry out content-based retrieval on the image database in modes such as draft drawing or framework adjustment so as to quickly find the needed image synthesis and creating resources.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method and system for semi-automatically constructing an image database including object classification and fine contour segmentation and annotation. Background technique [0002] With the rapid development of multimedia technology, various image and video resources are extremely rich, and a large number of images shot and created by users are added to the Internet every day. However, these images often only contain text labels, and the accuracy of the labeling is not guaranteed. Users often need to search for objects and characters that meet their wishes in a large number of images, and then search for them one by one in the massive results. [0003] In recent years, in order to solve this problem, researchers in the field of computer graphics and multimedia have done research on a large number of Internet image retrieval based on content. For example, a sketch-based ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F17/30G06T7/00
Inventor 胡事民陈韬马里千程明明
Owner TSINGHUA UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com